An investigation by Nikkei Asia highlights the growing scale of off-balance-sheet debt taken on by major US technology companies to fund the buildout of AI infrastructure. This financing, which does not appear directly in companies' published accounts, reportedly rose from roughly $1.7 trillion to $3.2 trillion over six months, a 90% increase.
This so-called hidden debt largely stems from complex financial arrangements: joint ventures with private equity firms, special purpose vehicles, and long-term lease agreements used to fund data center construction and chip purchases. These structures allow large tech firms to keep investing heavily in AI without directly inflating their balance sheets or weakening traditional financial ratios, even as they commit substantial sums over multi-year horizons.
The trend raises questions about the sustainability of the current surge in AI-related spending. Investment in data centers, semiconductors, and the energy needed to power them has reached unprecedented levels, driven by competition among players such as Microsoft, Meta, Google, Oracle, and Amazon to secure computing capacity. The growing reliance on external financing, often backed by uncertain future revenue streams, has drawn concern from some financial analysts about the systemic risk a slowdown in AI demand could create.
Beyond the headline figure, the finding feeds into a broader debate over the nature of the investment boom surrounding artificial intelligence. If revenue from AI services fails to keep pace with infrastructure spending, these off-balance-sheet commitments could turn into a vulnerability for the consolidated finances of the companies involved, as well as for their financial partners.